Oil Spill Scenarios in the Kotor Bay: Results from High Resolution Numerical Simulations
Total Page：16
File Type：pdf, Size：1020Kb
Journal of Marine Science and Engineering Article Oil Spill Scenarios in the Kotor Bay: Results from High Resolution Numerical Simulations Giulia Zanier 1,*,† , Massimiliano Palma 1,† , Andrea Petronio 1,† , Federico Roman 1,†, Vincenzo Armenio 2,† 1 Iefluids s.r.l., Piazzale Europa 1, 34127 Trieste, Italy; [email protected] (M.P.); a.petronio@iefluids.com (A.P.); f.roman@iefluids.com (F.R.) 2 Department of Engineering and Architecture, University of Trieste, 34127 Trieste, Italy; [email protected] * Correspondence: g.zanier@iefluids.com; Tel.: +39-040-558-3470 † These authors contributed equally to this work. Received: 2 January 2019; Accepted: 17 February 2019; Published: 25 February 2019 Abstract: A major threat for marine and coastal environment comes from oil spill accidents. Such events have a great impact on both the ecosystem and on the economy, and the risk increases over time due to increasing ship traffic in many sensitive areas. In recent years, numerical simulation of oil spills has become an affordable tool for the analysis of the risk and for the preparation of contingency plans. However, in coastal areas, the complexity of the bathymetry and of the orography requires an adequate resolution of sea and wind flows. For this reason, we present, to the best of the author’s knowledge, the first study on the subject adopting Large Eddy Simulations for both the low-atmosphere and sea dynamics in order to provide highly-resolved marine surface current and wind stress to the oil slick model, within a one-way coupling procedure. Such approach is applied to the relevant case of Kotor Bay (UNESCO heritage since 1979), in Montenegro, which is a semi-closed basin surrounded by mountains that is subject to an intense ship traffic for touristic purposes. Oil spill spots are tracked along ship paths, in two wind scenarios. Keywords: oil spill; numerical simulation; LES; low atmosphere; coastal flow; contingency plan; Kotor bay 1. Introduction Oil spill accidents represent a major threat to marine and coastal environment, impacting both biological species and human health, as well as economic, touristic and commercial activities. For example, according to data collected from 1977 to 2003 about 304,700 tons of oil have been released in the Mediterranean Sea mainly due to extensive marine traffic of oil tankers and ships [1,2]. Weather conditions, oil physical and chemical characteristics determine oil fate and persistence at sea. Most kinds of oils spread on the sea surface as a thin film, the slick is then driven by the sea currents and wind stress; furthermore, if the oil temperature drops below the pour-point, oil can solidify and form tar balls. In case of wavy and turbulent seas, small oil drops can detach from the oil slick and then, depending on their density, particles can either sink, float on the surface or be transported along the water column. Moreover, oil interacts with the surrounding environment. Immediately after spills, oil can evaporate and, under the action of wind and waves, it can absorb water droplets producing emulsion. Such phenomena, called weathering processes, change oil physical and chemical properties in time, strongly affecting oil fate and persistence at sea [3–6]. Given the significant impact of oil spill on the environment and economy of the area, over the years, efforts have been devoted to the preparation of contingency plans, aimed at ensuring fast J. Mar. Sci. Eng. 2019, 7, 54; doi:10.3390/jmse7020054 www.mdpi.com/journal/jmse J. Mar. Sci. Eng. 2019, 7, 54 2 of 23 response and to facilitate clean-up operations after accidents. In this context, oil spill numerical models have been widely established as helpful tools, both for development of contingency plans and for guiding clean-up operations. Oil slick models are usually integrated with hydrodynamic and meteorological models that provide sea currents and wind data. The modeling approaches can be classified as Lagrangian, Eulerian and Lagrangian/Eulerian hybrid models [4,7]. In the Lagrangian models, e.g., [2], oil slick is treated as a multitude of finite size particles, which are advected by a mean drift velocity plus a fluctuating turbulent component, the latter usually parametrized by means of a random walk technique. In the Eulerian method, e.g., [8], oil slick dynamics are derived from mass and momentum conservation equations. Finally, in hybrid Lagrangian/Eulerian models (see, for example [9]), a large number of particles parametrizes the oil slick immediately after the spill, and, as far as the width of the slick reaches a terminal value, the computation switches to a Eulerian model. In the present paper, we study the case of a hypothetical oil spill accident due to ship collision in Boka Kotorska Bay, a long and tortuous fjord situated in the Adriatic Sea. The study is aimed at preparation of contingency plans for the area under investigation. This area is under the UNESCO protection since 1979, for its own important natural and historical heritage. The prevention from possible hazardous situations is becoming urgent in light of the increased maritime traffic over the recent years. To make the study as realistic as possible, the typical ship path is considered within the bay [10,11] in conjunction with oil spill spots identified as dangerous from an environmental point of view. We use a novel approach to simulate oil slick dispersion in coastal areas characterized by surface currents and low atmosphere circulations governed by complex bathymetry, coastline and orography. We use a two-dimensional Eulerian model derived by Nihoul’s theory [12–14] for the oil slick. The oil slick simulation is coupled with LESCOAST [15,16], a high resolution hydrodynamic model used to simulate water circulation in coastal areas. Given the complex orography surrounding the bay, a preliminary low-atmosphere wind simulation is required to take into account the horizontal variability of wind stress. This latter is the main forcing item driving both the oil slick and the sea current in the upper layers, and it has to be properly modeled, for example as suggested in [17]. The paper is organized as follows: in Section2, we provide a brief overview of the hydrodynamic models for water and air domains; then, we introduce the oil spill model and finally we briefly describe the features of the area under investigation along with the boundary and initial condition for the simulations. Results of the most significant scenarios are reported and examined in Section3. Finally, the discussion is provided in Section4. 2. Materials and Methods In this section, we describe the methodology used for the study: in Section 2.1, we provide a brief description of the LESCOAST/LESAIR model used for the marine and low-atmosphere simulations; in Section 2.2, we present the oil spill model for the analysis of pollutant dispersion; in Section 2.3, we give a description of the Boka Kotorska Bay; in Section 2.4, we discuss the set-up of the simulations. 2.1. Hydrodynamical Model LESCOAST/LESAIR model [18,19] solves the filtered form of three-dimensional, non-hydrostatic Navier–Stokes equations under the Boussinesq approximation along with the transport equations for scalar quantities, i.e., salinity and temperature/humidity and temperature in marine/atmosphere simulations, respectively. The LESCOAST/LESAIR model uses a Large Eddy Simulation approach to parametrize turbulence, and the variables are filtered by a low-pass filter function represented by the size of the cells. The subgrid-scale fluxes (SGS), which come out from the filtering operation, are parametrized by a two-eddy viscosity anisotropic Smagorinsky model developed in [18]. Such method is effective in simulating coastal flows on sheet-like anisotropic computational grids. J. Mar. Sci. Eng. 2019, 7, 54 3 of 23 The complex geometry, which usually characterizes harbor and coastal areas, is treated using an Immersed Boundary Method (IBM), based on a direct forcing approach, as described in [20]; the technique is used to reproduce coastline, anthropogenic structures, bathymetry and topography. The filtered Boussinesq form of the Cartesian Navier–Stokes equations reads as follows: Continuity equation: ¶uj = 0, (1) ¶xj Momentum equation: 2 ¶ui ¶uiuj 1 ¶p ¶ ui Dr ¶tij + = − + n − 2eijkWjuk − gidi3 − , (2) ¶t ¶xj r0 ¶xi ¶xj¶xj r0 ¶xj Scalar transport equation: 2 s ¶s ¶ujs ¶ s ¶lj + = ks − , (3) ¶t ¶xj ¶xj¶xj ¶xj where ’−’ represents the filtering operation, ui represents the ith-component of the Cartesian velocity vector (u, v, w), xi represents the ith-component of the Cartesian coordinates (x, y, z), t is time, r0 is the reference density, p is the hydrodynamic pressure, n is the kinematic viscosity, eijk is the Levi–Civita tensor, Wi is the ith-component of the Earth rotation vector, Dr is the density anomaly, gi is the ith-component of the gravity vector, and tij is the SGS stress tensor which comes from the nonlinearity of the advective term, s is a scalar quantity (e.g., temperature and salinity/humidity), ks is scalar s diffusivity and lj is the SGS scalar flux. In the present low-atmosphere simulations, density variations are small and therefore buoyancy effects are neglected. In the marine simulations, we solve the transport equation of the scalar quantities, temperature T and salinity S, respectively. The fluid density is computed through the state equation: Dr r − r0 T S = = −b (T − T0) + b (S − S0), (4) r0 r0 T S where r0 is the reference density at the temperature T0 and salinity S0; b and b are respectively the coefficient of temperature expansion and haline contraction. At immersed boundaries, we apply the wall-layer model (IBWLM) presented in [21]; and at the open boundaries the Orlanski boundary condition is enforced [22], and it reads as: ¶ui ui + Ci = 0, (5) ¶t ¶xi where Ci is the phase velocity, calculated as the flux at the cell face.